CN114491772A

CN114491772A - Household layout generation method and device

Info

Publication number: CN114491772A
Application number: CN202210290430.6A
Authority: CN
Inventors: 刘念雄; 闫树睿; 苏航
Original assignee: Tsinghua University
Current assignee: Tsinghua University
Priority date: 2022-03-23
Filing date: 2022-03-23
Publication date: 2022-05-13
Anticipated expiration: 2042-03-23
Also published as: CN114491772B; US20230306165A1

Abstract

The invention relates to a house type layout generation method and device. The method comprises the following steps: acquiring a current house type state; room information of one or more rooms to be deployed is obtained through a particle swarm algorithm, wherein the room information comprises the length and the width of the room; each particle in the particle swarm represents size information of a room to be deployed; performing multiple Monte Carlo tree searches according to the room information to obtain coordinate information of the one or more rooms to be deployed; comprehensively scoring the room information and the coordinate information through an evaluation function to obtain a scoring result; and calculating the rated times, and deploying the one or more rooms to be deployed according to the highest scoring result in the scoring results. By the method provided by the invention, the quality of the generated result can be ensured, diversified design conditions can be met, the calculation time is saved, and the efficiency is greatly improved.

Description

Household layout generation method and device

Technical Field

The invention relates to the technical field of intelligent design, in particular to a house type layout generation method and device.

Background

House design is a relatively professional and complex task, and usually requires a designer to receive years of design training and a certain accumulation of experience, and a floor layout design needs to be performed by combining self professional knowledge and design experience. During the design process, the designer needs to manually draw the room areas, such as the shape and position relationship of the areas of a restaurant, a living room, a bedroom, a toilet and the like in a house-type map, so that a method is needed to assist the designer or non-professional person in the automatic house-type room arrangement.

In recent years, many studies have been made to generate a residential dwelling plane through a generative countermeasure network in deep learning. Generative confrontation networks were proposed by Goodfellow in 2014. In 2017 Isola P et al proposed pixel2 pixels, and GAN was used to create building facades. Huang et al used GAN to generate flat pictures of residential housing in 2018. Subsequently, Nautata N et al propose House GAN and House GAN + +, and generate a House type layout with the room relationship "graph" as the generation condition.

The GAN is mainly structured by two neural networks, a generator network (generator) and a discriminant network (discriminator). The generation network is responsible for generating target data, and the discrimination network is responsible for distinguishing the generated data from the real data. The training process is to enable the generation network and the discrimination network to play games, the optimization goal of the generation network is to enable the generation data distribution to approach the real data distribution continuously, and the optimization goal of the discrimination network is to distinguish the generation data from the real data to the maximum extent. Due to the existence of the discrimination model, the generated network can well learn to approach to the real data distribution on the premise of no large amount of prior knowledge, and finally the generated result achieves the effect of being false or spurious.

However, training of the generative confrontation network often requires a large amount of data, and under the condition of limited data quantity, the generalization capability of the model is often influenced. Meanwhile, the model learns the characteristics of the data, so that the quality of the data determines the training effect of the model to some extent. For residential dwellings, training based solely on these dwellings can affect the quality of the results generated, since the dwellings themselves on the market have a good or bad score. Meanwhile, because the requirements of different regions on the house design are different, the relevant design specifications can be changed along with the change of the times, and the model trained based on specific data is difficult to adapt to the differences and changes.

Therefore, a new method is needed to explore a reasonable residential layout by computer aided designers or non-professionals for automated residential room layout.

Disclosure of Invention

In view of this, the present application provides a method and an apparatus for generating a house layout.

In a first aspect, the present application provides a house layout generating method, including:

acquiring a current house type state;

room information of one or more rooms to be deployed is obtained through a particle swarm algorithm, wherein the room information comprises the length and the width of the room; each particle in the particle swarm represents size information of a room to be deployed;

performing multiple Monte Carlo tree searches according to the room information to obtain coordinate information of the one or more rooms to be deployed;

comprehensively scoring the room information and the coordinate information through an evaluation function to obtain a scoring result;

and calculating the rated times, and deploying the one or more rooms to be deployed according to the highest scoring result in the scoring results.

Preferably, the acquiring the current house type state includes:

obtaining design conditions of the house type, wherein the design conditions comprise the boundary range of the house type, the position of a main entrance, a lighting surface and adjacent building information, and determining the current house type state according to the design conditions.

Preferably, the room information further includes spatial information between rooms, the spatial information between the rooms including: and adjacent, separated and intersected, and representing the spatial information between the rooms through a room adjacency matrix.

Preferably, the house type state is expressed in a form of a matrix, which is

The method comprises the steps of acquiring a house type condition matrix, acquiring a house type design boundary, acquiring a long size information and a wide size information, wherein env represents the house type condition matrix, e represents the house type design boundary, x represents an x-axis coordinate, y represents a y-axis coordinate, w represents a long size information, and d represents a wide size information.

Preferably, the acquiring, by a particle swarm algorithm, room information of one or more rooms to be deployed includes:

obtaining different sizes of the room through continuous iteration of the particle positions, wherein the j-th particle position is calculated through the following formula

Where x denotes the position of the particle, ω denotes the inertia factor, v denotes the velocity of the particle, c₁,c₂Characterizing the learning factor, rand characterizing random numbers subject to [0,1) uniform distribution,

characterizing the historical best position, gbest, of particle i at the j-th calculation_iThe global optimal position of the particle i at the j-th calculation is characterized.

Preferably, the performing the plurality of monte carlo tree searches according to the room information includes:

constructing a current search graph according to the current house type state, wherein a node of each search graph represents a deployment condition of a room to be deployed;

and determining room information of a current room to be deployed in the one or more rooms to be deployed, and performing operation on the current search graph for a plurality of times through Monte Carlo tree search to obtain a search result, wherein the search result comprises an optimal node of the current search graph.

Preferably, each of the operations performed on the current search graph by the monte carlo tree search includes:

selecting a node as a root node according to a current search graph, and expanding the root node based on the root node to obtain a plurality of first child nodes;

screening the plurality of first sub-nodes by a pruning method to obtain a plurality of second sub-nodes, wherein the second sub-nodes represent actual expandable nodes after pruning;

performing random search based on the second child node, and evaluating a result obtained after the random search through an evaluation function;

and updating the weight of the relative node in the search graph according to the evaluation function score, wherein the optimal node is the node with the maximum weight.

In a second aspect, the present application provides a house layout generating apparatus, including:

the first acquisition module is used for acquiring the current house type state;

the second acquisition module is used for acquiring room information of one or more rooms to be deployed through a particle swarm algorithm, wherein the room information comprises the length and the width of the room; each particle in the particle swarm represents size information of a room to be deployed;

the third acquisition module is used for carrying out multiple Monte Carlo tree searches according to the room information to obtain the coordinate information of the one or more rooms to be deployed;

the evaluation module is used for carrying out comprehensive scoring on the room information and the coordinate information through an evaluation function to obtain a scoring result;

and the deployment module is used for calculating the rated times and deploying the one or more rooms to be deployed according to the highest scoring result in the scoring results.

In a third aspect, the present application provides a computing device comprising a processor and a memory, wherein the memory has stored therein computer program instructions, which when executed by the processor, perform the method according to any of the embodiments of the first aspect.

In a fourth aspect, the present application provides a computer-readable storage medium comprising computer-readable instructions which, when read and executed by a computer, cause the computer to perform the method according to any of the embodiments of the first aspect.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

Fig. 1 is a schematic view of an application scenario of the technical solution provided in the embodiment of the present application;

fig. 2 is a schematic diagram of a house type generation process provided in an embodiment of the present application;

fig. 3 is a schematic flow chart of a house layout generation method according to an embodiment of the present application;

fig. 4 is a schematic step diagram of a house layout generation method according to an embodiment of the present application;

fig. 5 is a schematic diagram of a house-type vectorization method provided in an embodiment of the present application;

FIG. 6 is a schematic view of a house type disassembly according to an embodiment of the present disclosure;

FIG. 7 is a schematic representation of spatial relationship of house types according to an embodiment of the present application;

fig. 8 is a schematic diagram of a house layout generating apparatus according to an embodiment of the present application;

fig. 9 shows a schematic structural diagram of a computer device provided in an embodiment of the present specification.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

For the convenience of understanding of the embodiments of the present invention, the following description will be further explained with reference to specific embodiments, which are not to be construed as limiting the embodiments of the present invention.

Fig. 1 is a schematic view of an application scenario of the technical solution provided in the embodiment of the present application. As shown in figure 1, by the technical scheme of the application, the buildings can be reasonably arranged on a house type plan of the existing buildings. In the present application, the description will be given mainly taking the layout of a house as an example. It will be appreciated that the dwelling may equally be replaced with other spatial arrangements in various possible business scenarios, such as cargo placement, city design, interior design, strategy games, etc.

Fig. 2 is a schematic diagram of a house type generation process provided by an embodiment of the present application, and as shown in fig. 2, under an environmental condition of a given house type, rooms to be deployed in a room are combined in an ideal manner. As can be seen from fig. 2, the method of the present application employs a Particle Swarm Optimization (PSO) and a Monte-Carlo Tree Search (MCTS).

PSO is a biomimetic algorithm that simulates the birds in a flock of birds by means of particles. In the combinatorial optimization problem, different combinations between variables are treated as individual particles, which have two properties: speed and position. In the iteration process, the particles update the position information of the next step according to the speed of the particles. And each particle independently searches an optimal solution in a search space and records the optimal solution as a current individual extreme value, the individual extreme value is shared with other particles in the whole particle swarm, a current global extreme value can be obtained through comparison, and the speed and the position of each particle in the particle swarm are adjusted according to the current individual extreme value found by each particle and the current global extreme value shared by the whole particle swarm. And finally, continuously iterating to realize optimization of variable combinations. The PSO algorithm has the characteristics of simplicity, convergence rate block and capability of parallel computation.

The monte carlo tree search is used to solve the optimal decision problem. Through the Monte Carlo tree search, the optimal decision of the room placement position and size can be realized. The monte carlo tree search may be divided into four steps. Selection, expansion, simulation, and back propagation. First, in the selection phase, a node N that needs to be expanded needs to be selected downward from the root node. Secondly, according to the action which is not expanded in the node N, an N child node is created in the search tree. Third, to get an initial score for the child node. It is necessary to start with the child node and let the game randomly proceed until a game outcome is obtained, which will be used as the initial score for the child node. Finally, after the simulation of the child node is finished, the parent node N and all nodes on the path from the root node to N update the accumulated weight of each child through the UCT equation according to the result of the simulation. The search tree is expanded every iteration, and the scale of the search tree is continuously increased along with the increase of the number of iterations. And when the simulation times reach a set value, stopping simulation, and selecting the best child node under the root node as the decision result.

Referring to fig. 2, the layout of the house type is completed step by step, the size of each room to be deployed in the house type is determined by a particle swarm algorithm, and then the coordinates of the rooms are obtained by the room size through a monte carlo tree search method. And continuously optimizing the house type layout through multiple iterations to obtain the finally optimized house type layout.

Fig. 3 is a schematic flow chart of a house layout generation method according to an embodiment of the present application. As shown in fig. 3, firstly, determining design conditions, determining a house type state according to the design conditions of the house type, then obtaining size information of a room to be deployed through a PSO algorithm, determining coordinates of the room through the size information of the room by an MCTS method, and scoring the size information and the coordinate information of the room through an evaluation function; and if the iteration step number reaches the set iteration step number, deploying the room according to the highest score in the scores to obtain the final house type, and if the iteration step number does not reach the set iteration step number, returning to the PSO method for the next iteration.

Fig. 4 is a schematic step diagram of a house layout generation method according to an embodiment of the present application. Fig. 4 is further explained below in conjunction with fig. 2 and 3. The house type layout generating method of the application can comprise the following steps:

step S401: and acquiring the current house type state.

A single dwelling-type space unit can be divided into two categories, an interior room and an exterior environment. The interior room refers to all closed rooms (such as bedrooms, toilets and the like) and open spaces (such as living rooms, restaurants and the like) except for the aisle. Fig. 5 is a schematic diagram of a house-type vectorization method provided in the embodiment of the present application, and fig. 6 is a schematic diagram of a house-type dismantling provided in the embodiment of the present application. Referring to fig. 5 and 6, as shown in fig. 5a, a house type room includes two types: closed rooms (e.g., bedrooms, toilets, etc.) and open rooms (e.g., living rooms, restaurants, etc.). These room combinations can be abstracted as combinations of rectangles, and the information of the rooms can be described by size and coordinates. As shown in FIG. 6, most rooms are in an "adjacent" relationship, such as rooms A and B in FIG. 6B, but some rooms are in an "intersecting" relationship, such as rooms B and C in FIG. 6B. In an intersecting relationship, as shown in room B of fig. 6a, a partial "intrusion" of one room into another room, where the "intruded" room plane is in the form of a concave polygon, but can still be viewed as a complete rectangle by way of padding. The combination between all rooms in one house-type plane can be regarded as a combination of rectangles. From the abstracted rectangular room, the room form can be quantified, as shown in fig. 5c, which includes room position coordinates and the size of the room, including the width and depth of the room. The size refers to the length and width of each room rectangular unit, and the room is too large or too small to be used.

In the present application, all room unit planes and external environment elements are abstracted to rectangles and are represented by coordinates and dimensions of length and width. Thus, the entire house layout can be described by the set of x-axis, y-axis coordinates and length and width dimensions. It should be noted that abstracting the floor plan to a rectangle is a simplification, and the computational difficulty can be reduced by this simplification. In the case of buildings, although the plane of the rooms in the building has an irregular form, most of the rooms are rectangular planes, so that most of the house layouts can be represented by the method even if all the rooms are simplified into rectangles.

Specifically, the house type state is expressed in the form of a matrix, which is

In some more specific embodiments, the obtaining the current house type state includes:

The external environment of a house type space unit refers to four aspects of boundary range, lighting surface, entrance and adjacent buildings. A house layout is a combination of room units and room units with the outside environment (house design boundary).

Step S402: acquiring room information of one or more rooms to be deployed through a particle swarm algorithm, wherein the room information comprises the length and the width of the room; each particle in the population of particles includes dimensional information for one room to be deployed.

The constraint condition of the room size is often a continuous integer interval, the constraint condition is less, and the room size particles can be quickly generated through a PSO algorithm.

According to the house type condition matrix, particles containing length and width dimension information of the room to be deployed can be generated, and the room to be deployed is formed by a plurality of pieces of room information to be deployed and is expressed as a room matrix to be deployed

The size represents a room matrix to be deployed, r represents a room to be deployed, w represents width information, d represents length information, P is a PSO algorithm, and env represents a house type condition matrix.

In some possible embodiments, the room information further includes spatial information between the rooms, the spatial information between the rooms including: and adjacent, separated and intersected, and representing the spatial information between the rooms through a room adjacency matrix.

Fig. 7 is a schematic diagram illustrating a spatial relationship expression of house types according to an embodiment of the present application. Referring to fig. 7, the spatial relationship of the present application refers to the adjacent relationship between interior rooms, and between interior rooms and the external environment. The original house type is planed into a plane diagram, the room relation of different rooms in the plane diagram forms a room space relation diagram, and the room space relation is described through a room adjacency matrix. The relationship between rooms can be summarized into three relationships of separated, adjacent and crossed.

In some possible embodiments, the obtaining room information of one or more rooms to be deployed by a particle swarm algorithm comprises:

In the present application, the house type room size is controlled by modulus, so all room sizes are multiples of one fixed size unit, so the room size is encoded according to discrete integer variables by using a discrete binary particle swarm algorithm.

Since the particle x is a binary vector, the method of updating the particle x in an iterative manner is bit reversal. When the speed of the particle is larger, the difference between the particle and the optimal particle is larger, so that the probability of turning is larger; conversely, the smaller the velocity, the smaller the gap between the representation and the current optimal particle, and therefore the smaller the probability of flipping.

In a more specific example, the bit flip expression is

Wherein x denotes the position of the particle, ω denotes the inertia factor, v denotes the velocity of the particle, c₁,c₂Characterizing the learning factor, rand characterizing random numbers subject to [0,1) uniform distribution,

characterizing the historical best position, gbest, of particle i at the j-th calculation_iThe global best position of the token particle i at the j-th calculation, k token index,

the mapped value of the position velocity of k index at step j for particle i,

is the value of the index position k of particle i at step j.

Step S403: and performing multiple Monte Carlo tree searches according to the room information to obtain the coordinate information of the one or more rooms to be deployed.

After the room size is determined, in order to eliminate a large number of useless solutions and compress a search space, the room coordinates need to be screened more finely, so that the coordinates are more suitable for pruning through an MCTS algorithm, and the advantage of tree search accuracy is exerted. And simultaneously, in the MCTS searching stage, parallel calculation is carried out by taking the particles generated by the PSO as units so as to further save time.

After the room size is determined, the deployment position of the internal room is searched through MCTS, and a room matrix composed of x-axis and y-axis coordinates is obtained and is represented as

Wherein coords represents a room matrix, r represents a room, x represents an x-axis coordinate, y represents a y-axis coordinate, M is an MCTS method, env represents a house type condition matrix, and size represents a room matrix.

In some more specific embodiments, performing multiple monte carlo tree searches according to the room information includes:

and constructing a current search graph according to the current house type state, wherein the node of each search graph represents a deployment condition of a room to be deployed.

Specifically, the current house type state is determined according to the obtained boundary range, entrance position, lighting surface and adjacent building information of the current house type, and a search graph is constructed according to the current house type state.

In some more specific embodiments, each of the plurality of operations performed on the current search graph by the monte carlo tree search comprises:

step A: selecting a node as a root node according to a current search graph, and expanding the root node based on the root node to obtain a plurality of first child nodes;

specifically, each room to be deployed has a size interval (width interval and length interval), all combinations of the sizes and the coordinates are obtained according to the coordinates of each room, and each node corresponds to a possible deployment situation of one room to be deployed, that is, each node corresponds to a combination of a set of sizes and coordinates of one room to be deployed. For example, the room to be deployed currently is the master bedroom, and meanwhile, the layout of the current house type is not started, so that the current house type state is the root node. And starting to layout the master bedroom based on the current house type state, so that the house type states of a plurality of layout master bedrooms can be obtained, and each state in the house type states of the plurality of layout master bedrooms is a first node.

And B: screening the plurality of first sub-nodes by a pruning method to obtain a plurality of second sub-nodes, wherein the second sub-nodes represent actual expandable nodes after pruning;

in a more specific example, the screening of the first plurality of child nodes by pruning includes position pruning and size pruning.

In order to eliminate unnecessary options, the exploration space is compressed, and the family search graph needs to be pruned. It will be appreciated that pruning can compress the exploration space by excluding some obviously unsuitable room forms.

For example, if the current house type state when the current house type does not start to be laid out is selected as a root node, and the layout is expanded, the house type state of the deployed master bedroom is a first child node, in some layout situations, the layout of the master bedroom cannot well meet normal use requirements, the layout which does not meet the use requirements is screened out, and a plurality of actual extensible house type states after pruning are obtained and are second child nodes. If one layout in the plurality of pruned actual expandable house type states is selected as a root node, the layout is expanded, namely, a next room to be deployed, such as a guest bedroom, is deployed to obtain a plurality of guest bedroom layout states after the host bedroom is deployed, the guest bedroom layout states are taken as first child nodes, the guest bedroom layout states after the host bedroom is deployed are pruned, and the obtained result is taken as second child nodes.

In the present invention, all room forms are represented by rectangles. Although the plane of the building room has an irregular form, most rooms are rectangular planes, so that representing all rooms in the house type by rectangles can realize description of most of the layout of the building plane, and simultaneously simplifies the calculation difficulty, so that the model is balanced in generalization capability and calculation efficiency.

Position pruning requires defining spatial relationships (e.g., intersection, adjacency) between each room first, and pruning according to the spatial relationships. And the size pruning is to judge whether the area interval is met or not and prune the unsatisfied branch. The position pruning can eliminate unnecessary coordinate points for placing the room, and the size pruning can further control the size of the room on the premise of determining the position of the room. All feasible situations of next room placement under a specific house type state can be obtained through pruning.

In a more specific example, the specific formula for position pruning is:

in the formula (I), the compound is shown in the specification,

characterize the set of coordinates in which the ith room can be placed,

characterize placeable points of rooms adjacent to the ith room,

the remaining empty area when characterizing the ith room placement can place points,

characterizing placeable point x-axis and y-axis coordinates that require rooms adjacent to the ith room,

x-axis and y-axis coordinates of placeable points that characterize the remaining empty area at the time of the ith room placement.

Because the relationship between the rooms is defined in advance, all the deployment conditions of the rooms to be deployed with proper position information can be obtained through the intersection of the blank area when the rooms to be deployed are placed and the placeable points of the rooms adjacent to the rooms to be deployed.

For example, as shown in fig. 6, it is defined in advance that the B room and the C room intersect, and a plurality of possible deployment situations of the B room can be determined according to the intersection of the previously defined spatial relationship and the points of the non-arranged areas inside the current house.

In a more specific example, the specific formula for size pruning is:

in the formula (I), the compound is shown in the specification,

characterize the set of rooms in which the ith room can be placed,

characterize the set of coordinates in which the ith room can be placed,

a set of widths characterizing the ith room,

the depth set of the ith room is characterized.

And multiplying the coordinate set and the size set of the room to be deployed to obtain a room set with a proper size of the room to be deployed.

And C: and performing random search based on the second child node, and evaluating a result obtained after the random search through an evaluation function.

In some possible embodiments, the evaluating the result obtained after the random search by the evaluation function includes evaluating the number of rooms, evaluating the intersection area, evaluating the aisle, wherein the aisle evaluation includes evaluating the aisle area, evaluating the aisle and the room adjacent condition, and evaluating the aisle width.

Pruning is a strong constraint on the house type feasible solution, while the merit function is a weak constraint on the feasible solution. By scoring each final result and updating the weight of each node of the search graph according to the score, the evaluation function can guide the algorithm to search towards the optimal strategy direction.

In a more specific example, the specific formula of the evaluation function is:

max Score＝S_n+S_o+S_P

in the formula, Score represents the house type total Score, S_nCharacterizing the number of rooms score, S_oCharacterization intersection area score, S_PThe aisle score is characterized.

In a more specific example, the specific formula for the room number score is:

in the formula, n_dCharacterizing the number of rooms already deployed, n_tThe total number of rooms in the planned deployment is characterized.

The specific formula of the intersection area score is as follows:

in the formula (I), the compound is shown in the specification,

the total area of all the rooms that have been deployed is characterized,

characterizing the total area, s, of the intersection of all deployed room shapes_bThe boundary extent area is characterized.

The specific formula of the aisle score is as follows:

S_P＝P_o+P_w+P_a

in the formula, P_wCharacterizing aisle width score, P_aCharacterizing aisle area score, P_dThe characterization aisle and room adjacency score.

In a more specific example, the aisle width score is expressed as:

in the formula, w_stdThe target minimum width of the aisle is characterized, and w represents the current minimum width of the aisle.

The aisle area score is expressed as:

in the formula, a_stdCharacterizing the minimum area of the aisle target, and a characterizing the current area of the aisle.

The aisle-to-room adjacency score is expressed as:

in the formula (d)_iThe distance of the crossing from the ith room is characterized.

Step D: and updating the weight of the relative node in the search graph according to the evaluation function score, wherein the optimal node is the node with the maximum weight.

In a more specific example, a specific formula for updating the weights of the relative nodes in the search graph is as follows:

wherein W represents the weight of a node, n_jCharacterizing the number of times a child node is accessed, q_jThe sub-nodes are characterized in that,

characterizing child nodes q_jThe value in the i simulations, i represents the simulation times, n represents the total number of times the root node is visited, and c represents the hyper-parameter.

Wherein the hyperparameter c is a constant and the theoretical value is

The larger the c value, the more preferable the breadth search, and the smaller the c value, the more preferable the depth search.

Step S404: comprehensively scoring the room information and the coordinate information through an evaluation function to obtain a scoring result;

and after the room deployment is finished, calculating the house type fitness through the evaluation function E. Finally, the optimal size and coordinate combination sizes of the internal room are obtained by updating the fitness of the maximized sizes and coords of the sizes parameters^*,coords^*Is shown as

Wherein size denotes the optimal size combination (matrix), coords denotes the optimal coordinate combination (matrix), arg denotes the function, E denotes the rating function.

Step S405: and calculating the rated times, and deploying the one or more rooms to be deployed according to the highest scoring result in the scoring results.

And if the iteration step number reaches the set iteration step number, deploying the room according to the highest score in the scores to obtain the final house type, and if the iteration step number does not reach the set iteration step number, returning to the PSO method for the next iteration.

The Monte Carlo tree search and particle swarm optimization are combined, the method is used for automatic design of residential house layout, design assistance is provided for architects, and non-professional personnel can be helped to achieve design intention or design evaluation. In the combined optimization process of the house type rooms, the model can prune the nodes through an MCTS algorithm to compress a search space, so that an ideal house type layout result is obtained while the generalization capability of the model is ensured. Meanwhile, the calculation time is saved by utilizing the characteristics of high convergence speed and parallel calculation of the PSO algorithm, and the effect and the efficiency are considered. Experimental results show that the results of generating the layout of the user type by the PSO-MCTS are superior to those of the single MCTS and PSO algorithms, and the calculation efficiency is well balanced. The method can deal with diversified design conditions while ensuring the quality of the generated result, is not limited by case data compared with a deep learning method, can obtain a result superior to the case data, and can save the calculation time and greatly improve the efficiency compared with a simple MCTS method.

Based on the house type layout generating method provided in the above embodiments, a house type layout generating device is provided in this embodiment, and specifically, fig. 8 shows an alternative structural block diagram of the house type layout generating device, the device is divided into one or more program modules, and the one or more program modules are stored in a storage medium and executed by one or more processors, so as to complete the present invention. The program module referred to in the present invention refers to a series of computer program instruction segments capable of performing specific functions, and is more suitable for describing the execution process of the layout generation than the program itself, and the following description will specifically describe the functions of each program module in this embodiment. The device specifically includes:

the first acquisition module is used for acquiring the current house type state.

The second acquisition module is used for acquiring room information of one or more rooms to be deployed through a particle swarm algorithm, wherein the room information comprises the length and the width of the room; each particle in the population of particles characterizes dimensional information of one room to be deployed.

And the third acquisition module is used for carrying out multiple Monte Carlo tree searches according to the room information to obtain the coordinate information of the one or more rooms to be deployed.

And the evaluation module is used for carrying out comprehensive scoring on the room information and the coordinate information through an evaluation function to obtain a scoring result.

Fig. 9 shows a schematic structural diagram of a computer device provided in an embodiment of the present specification, where the computer device may include: a processor 910, a memory 920, an input/output interface 930, a communication interface 940, and a bus 950. Wherein the processor 910, the memory 920, the input/output interface 930, and the communication interface 940 are communicatively coupled to each other within the device via a bus 950. The computer device may be configured to perform the method illustrated in fig. 2, as previously described.

The processor 910 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an Application Specific Integrated Circuit (ASIC), or one or more Integrated circuits, and is configured to execute related programs to implement the technical solutions provided in the embodiments of the present disclosure.

The Memory 920 may be implemented in the form of a ROM (Read Only Memory), a RAM (Random Access Memory), a static storage device, a dynamic storage device, or the like. The memory 920 may store an operating system and other application programs, and when the technical solution provided by the embodiments of the present specification is implemented by software or firmware, the relevant program codes are stored in the memory 920 and called by the processor 910 to be executed.

The input/output interface 930 is used for connecting an input/output module to realize information input and output. The i/o modules may be provided as components within the device (not shown) or may be external to the device to provide corresponding functionality. The input devices may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output devices may include a display, a speaker, a vibrator, an indicator light, etc.

The communication interface 940 is used for connecting a communication module (not shown in the figure) to implement communication interaction between the present device and other devices. The communication module can realize communication in a wired mode (such as USB, network cable and the like) and also can realize communication in a wireless mode (such as mobile network, WIFI, Bluetooth and the like).

Bus 950 includes a pathway to transfer information between various components of the device, such as processor 910, memory 920, input/output interface 930, and communication interface 940.

It should be noted that although the above-mentioned device only shows the processor 910, the memory 920, the input/output interface 930, the communication interface 940 and the bus 950, in a specific implementation, the device may also include other components necessary for normal operation. In addition, those skilled in the art will appreciate that the above-described apparatus may also include only those components necessary to implement the embodiments of the present description, and not necessarily all of the components shown in the figures.

It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.

It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program instructing relevant hardware, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.

The above embodiments are provided to further explain the objects, technical solutions and advantages of the present invention in detail, it should be understood that the above embodiments are merely exemplary embodiments of the present invention and are not intended to limit the scope of the present invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.

Claims

1. A house type layout generation method is characterized by comprising the following steps:

acquiring a current house type state;

acquiring room information of one or more rooms to be deployed through a particle swarm algorithm, wherein the room information comprises the length and the width of the room; each particle in the particle swarm represents size information of a room to be deployed;

2. The method of claim 1, wherein obtaining the current subscriber type status comprises:

3. The method of claim 1, wherein the room information further comprises inter-room spatial information, the inter-room spatial information comprising: and adjacent, separated and intersected, and representing the spatial information between the rooms through a room adjacency matrix.

4. The method of claim 1, wherein the house type state is represented in a matrix form as

5. The method according to claim 1, wherein the obtaining room information of one or more rooms to be deployed by a particle swarm algorithm comprises:

6. The method of claim 1, wherein performing a plurality of monte carlo tree searches based on the room information comprises:

constructing a current search graph according to the current house type state, wherein the node of each search graph represents a deployment situation of a room to be deployed;

7. The method of claim 6, wherein each of the operations performed on the current search graph by the Monte Carlo tree search comprises:

8. A house layout generating apparatus, comprising:

9. A computing device comprising a processor and a memory, wherein the memory has stored therein computer program instructions which, when executed by the processor, perform the method of any of claims 1-7.

10. A computer readable storage medium comprising computer readable instructions which, when read and executed by a computer, cause the computer to perform the method of any one of claims 1-7.